5 mistakes robotics startups should avoid

 

Years ago, robots were inflexible, hard to program, stationary and, except for industries with long production runs, cost-prohibitive. Plus, they were blind, deaf, incapable of tactile sensing, unintelligent and couldn’t be trusted to work next to people.

Now, thanks to stronger processing power, sophisticated software that simplifies programming and operation, machine learning (ML), machine vision (MV), more and better sensors and end effectors and other developments, robots are more flexible and capable than ever. In addition, the relative cost of robotics has gone down.

With the growing need for quality and throughput amidst a shortage of skilled workers, the use of robotics in manufacturing is expanding. New robotic applications are popping up in more industries including pharma and agriculture. Robots are also leaving the factory floor to do more domestic services and other tasks.

In its 2018 World Robotics Report, the International Federation of Robotics estimates the combined market value for 2019 through 2021 for non-manufacturing, professional-service robots to be $37 billion—almost tripling its previous value.

These new opportunities are attracting more pioneers of new robotic applications. Here are five pitfalls robotics startups need to avoid.

Developing software in a bubble

Many startups begin with developing software that will automate a production task. The robot is often an afterthought, a decision based on cost and availability.

Images of robots in factories often conjure up automotive assembly lines with robots easily working in synchronized precision. However, this is the result of many cumulative years of work from many engineers to coordinate and refine the process.

Startups should consider all components of the task in the beginning for easier integration at the end. This is especially true for companies that have no or little prior experience with robotics.

Underestimating prototype to production

In new product development, a concept goes from idea to prototype through the process of design, engineering and construction. A common mistake startups make is to underestimate what it takes to go from prototype creation to production.

Detailed engineering should make each deployment of the solution successful for the end user. The solution should be robust, reliable and easy to use. After the installation, a customer will need support, and startups should consider how to meet these customer-service needs.

Transitioning between these phases requires a major shift in the skills needed by the startup. The latter phase will take much more time and effort than many expect.

Not building an out-of-the-box solution

A startup’s automation offering must take into consideration the setup involved. The product should be simple to install and easy to use.

If a deployment requires engineers to be onsite for days making customized adjustments, the number of solutions will be limited by not just the speed of building systems, but by how many qualified engineers are available for installations in the field.

For startups that want to produce a system for thousands of locations, this requires an out-of-the-box solution.

Focusing on the technology, not the problem

An automation user typically does not care about how the solution works, whether the system uses machine learning or novel self-correcting algorithms, but instead if it works.

Startups often have trouble remembering this, often born out of their early years. When trying to attract venture capital, startups need to explain not only the problem solved, but also the cutting-edge technology used.

However, when offering customers a new variation of a pick-and-place robot, they need to know how well and how fast it picks and what happens if it jams, not about the elegance of the engineering. The focus should be on solving the users’ problem, which is what they are going to pay for.

Trying to reinvent the wheel

The worst thing a startup can do is waste time, money and effort on creating already commercially available products. For instance, creating custom motion-control software. Robot hardware and motion planning software have developed in symbiosis for the last 40 years, with the software optimizing the reliability and life of the robot to achieve the optimal ROI for a customer.

Another mistake is for startups to develop their own robot arms from scratch if one already exists and can do the task. Instead, take advantage of the decades-long knowledge embedded in currently available arms from larger robotics prodder providers.

Moreover, because of the investment size and ROI importance of industrial automation, an unknown system with unproven results offered by a newcomer can pose too much risk for the end user. Nobody wants to pay to be a guinea pig or a bad example that becomes a learning lesson for other companies.

Startups should consider working with well-established robotics partners to leverage their expertise and experience, but also their distribution systems and support networks and brand awareness to its own benefit.